Linked_Sensor_Data

Download Report

Transcript Linked_Sensor_Data

Publishing Linked Sensor Data
Semantic Sensor Networks Workshop 2010
In conjunction with the 9th International Semantic Web Conference
(ISWC 2010), 7-11 November 2010, Shanghai, China.
Presenter: Kerry Taylor, CSIRO ICT Centre, Canberra, Australia
Payam Barnaghi (Uni. of Surrey), Mirko Presser (Alexandra Institute), Klaus Moessner
(Uni. of Surrey)
Contact author: Payam Barnaghi ([email protected])
Centre for Communication Systems Research
University of Surrey
1
Sensor networks and accessing
physical world data
• There are currently ongoing research on creating large-scale
sensor/actuator networks; This will enable connecting millions of
devices that capture physical world data in a global scale.
• The sensors provide observation and measurement data from the
physical.
• The current data transmission on sensor networks mostly relies on
binary or syntactic data models which lack of providing machine
interpretable meanings to the data.
– Binary representation or in some cases XML-based data
– No general agreement
– Requires an pre-agreement on both communication parties to be able to
process and interpret the data
– Limited reasoning
– Limited interoperability
– Data integration and fusion issues
2
Physical world data on the Web
• The idea is providing sensor data on the same level as
the Web data.
– Semantic enrichment of data and integrating the real world data
into the digital world;
• Providing annotations and associating the descriptions to
existing ontologies and domain knowledge
• There are existing standards such as those provided by
OGC, SSN-XG Sensor Ontology,…
3
W3C SSN-XG ontology
SSN-XG annotations
makes observations
of this type
What it
measures
units
SSN-XG ontologies
where it is
4
4
Sensor ontologies and semantic data
• The ontologies and semantic models provide machineinterpretable descriptions
• There is no direct association to the domain knowledge
– What a sensor measures, where it is, etc.
– Association of an observation and/or measurement data to a
feature of interest.
• Including the domain knowledge and relating the
enriched description to the existing data in the digital
world will support semantic integration.
• Inference mechanism can process and analyse the
emerging semantics.
5
Semantic interoperability and semantic
integration
Making sensor-generated information usable as
a new and key source of knowledge will require
their integration into the (existing) information
space of Communities  Semantic Integration
6
Semantic integration- example
“I am a parcel
for Tom,
dropped once”
1010
1010
1010
1010
1010
1010
Middleware
“I am a Post
van, not
going to
Tom”
Middleware
“I am TWITTER”
Middleware
7
Semantic
Semantic
Middleware
Middleware
Middleware
Mash-up of
Real World
Knowledge
Semantic
Semantic integration- example
Description
Discovery
Integration
Distributed processing
8
Semantic integration
• Semantics allows to create reusable knowledge that
helps to
– understand who is talking to whom
– who is doing what
– and what the information means
• This enables the integration of information as
knowledge.
• On a large scale this machine interpretable data is a key
enabler and a necessity for the Real World Internet.
9
9
Publishing linked sensor data
• Using existing knowledge on the Web to annotate the
sensor resources.
• Associating sensor descriptions to the domain
knowledge.
• Defining links between sensor observation and
measurement and features of interests using the existing
knowledge and domain ontologies.
• Making sensor descriptions as a part of Web data and
accessible through standard interfaces.
10
Linked data principles
• The principles in designing the linked data are defined
as:
– using URI’s as names for things;
– using HTTP URI’s to enable people to look up those names;
– provide useful RDF information related to URI’s that are looked
up by machine or people;
– including RDF statements that link to other URI’s to enable
discovery of other related things of the web of data;
11
Linked Data- Connecting distributed data across
the Web
- There are more than 13.1 billion interlinked RDF triples.
- more than 142 million RDF links (properties).
12
12
Sensor data and linked data
13
* The middle layer is adapted from Amit Sheth et al., “Semantic Sensor Web”
Publishing linked sensor data
• We use existing linked-data to annotate sensor data and
to associate the description to the domain knowledge;
• We also publish the sensor data a linked data resources.
14
Using linked data for annotation
15
Using linked data for annotation –
location model
• We have a two layers location description;
• A detailed location ontology for local descriptions;
• A location attribute (concept) obtained from linked data
(e.g. DBPedia, GeoNames);
• The local ontology provides detailed location description
(e.g. rooms, buildings on our campus) and the linked
data concepts provide high-level concepts (e.g.
University of Surrey) and then we linked these two
models;
16
Using linked data for annotation –
location model
Internal location
ontology (local)
Lined-data location
(external) 17
Using and reasoning the publishing
linked sensor data
18
Components and architecture
19
The World at your Fingertips
The world is the knowledge
base
Despite data volume, heterogeneity, distribution, dynamics:
Integration/access all that data like a set of interconnected
resources in an information network!
- Structured Querying
- Integrated Views
- Aggregation, Analyses  Reasoning upon the data
20
20
Thank you!
21